Thin Solid Films 517 (2009) 63236326

Contents lists available at ScienceDirect

Thin Solid Films

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f

Solgel synthesis, comparative characterisation, and reliability analyses of undoped and Al-doped zinc oxide thin lms
Mehmet opurolu a, Shane O'Brien a,, Gabriel M. Crean a,b
a b

Tyndall National Institute, Cork, Ireland Department of Microelectronic Engineering, University College Cork, Ireland

a r t i c l e

i n f o

a b s t r a c t
Thin lms of solgel derived undoped and Al-doped zinc oxide material systems were fabricated for potential use in transparent conducting oxide modules. A comparative characterisation of the functional properties of the lms was performed. Undoped zinc oxide thin lms were obtained with desired properties. Al doping reduced the average crystallite size, and led to a denser and less porous morphology, and also caused an increase in transparency in the UV region. An improvement in electrical conductivity was achieved upon Al doping. Temperature cycling and accelerated delamination test results demonstrated the high physical reliability of the thin lms of the material system obtained in this study. 2009 Elsevier B.V. All rights reserved.

Available online 11 March 2009 Keywords: Zinc oxide Solgel method Doping Reliability

1. Introduction Transparent conducting oxides (TCOs) are widely used as electrodes on at panel displays, sensors and solar cells; and to-date, indium tin oxide (ITO) has been the main material of choice [1]. However, a replacement for ITO is now required due to both cost and environmental issues. Both undoped and doped solgel derived zinc oxide is now attracting signicant attention due to its wide range of electrical and optical properties [29]. In addition to characterisation of the functional properties, assessment of the physical reliability of optoelectronic devices is also of signicant importance from application point of view. Temperature cycling tests in appropriate environmental laboratory test chambers, for instance, can be devised simulating the particular environmental conditions as described by the Generic Reliability Assurance protocol [10]. An accelerated delamination test can also be employed to provide a rapid evaluation of the susceptibility of a TCO [11]. The latter technique is based on stressing the TCO by applying heat, humidity, and a direct current (DC) bias to drive sodium ions to the TCOglass interface to predict the probability of thin lm TCO module failure via electrochemical delamination of the TCO. In this study, functional properties of the solgel derived undoped and Al-doped zinc oxide lms were comparatively characterised by means of X-ray diffractometry (XRD), scanning electron microscopy (SEM), UV/vis spectrophotometry, and electrical resistivity measurements, and their reliability analyses were performed in terms of temperature cycling and accelerated delamination tests.

2. Experimental details 2.1. Synthesis Zn(C2H3O2)22H2O (Sigma-Aldrich) was dissolved in the mixture of isopropanol (IPA; Rockwood, isoclean), as a solvent, and 2-aminoethanol (MEA; Sigma-Aldrich), as both a complexing agent and a base, while stirring at 60 C for 1 h. The molar ratio of Zn to MEA was kept constant at 1. Then, an adequate amount of IPA was added so as to attain a sol of [Zn]= 0.7 M. In the case of doping, a 0.2 M ethanolic Al solution was rst prepared by dissolving the required amount of Al(NO3)39H2O (Aldrich) in absolute ethanol (Aldrich). This solution was then added to the sol directly after the dissolution of Zn(C2H3O2)22H2O in IPA + MEA, in such a way that the resultant Al concentration in the mixture was 5 mol% ([Al]/([Al] + [Zn]) 100). In this case, the molar ratio of total metals (Zn + Al) to MEA was kept constant at 1. An adequate amount of IPA was then added so as to attain a sol of [Zn]+ [Al] = 0.7 M. All sols, undoped or doped, were left to age for 1 day at ambient temperature in a closed vessel. The overall outline of the synthesis is given in Fig. 1.

2.2. Film preparation and processing Following the sol ageing, the mixtures were spin-coated onto sodalime glass slides (Menzel-Glser, with a Na2O content of ~14.30 mass%) using a WS-400A-6NPP/LITE Spin-Coater (Laurell Technologies) at a spin rate of 3000 rpm for 20 s. The coated slides were then transferred into a laboratory furnace (Heraeus Instruments, M104), which had been set to 275 C. After 10 min, the furnace was heated to 550 C, using an approximate heating rate of 10 C min 1. Following an annealing of 1 h, the furnace was left to cool down to ambient temperature. The lm preparation and processing are outlined in Fig. 2.

Corresponding author. E-mail address: shane.obrien@tyndall.ie (S. O'Brien). 0040-6090/$ see front matter 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2009.02.056

6324

M. opurolu et al. / Thin Solid Films 517 (2009) 63236326

In order to quantify the intensity ratio of the (002) crystal orientation, (002), was dened as: 002 = I002 I002 + I100 + I101

where I(002), I(100), and I(101) denoted the intensities of (002), (100), and (101) crystal orientations, respectively. Top-view and cross-sectional SEM images of the lms were collected using a Hitachi S-4000 Field Effect SEM instrument, whose operating voltage was 20 keV. Film thickness was estimated from the cross-sectional SEM images. UV/vis spectroscopy was performed with a Shimadzu 2401 PC UV/vis spectrophotometer. Electrical resistivity measurements were carried out using a Keithley 617 electrometer evaporating two coplanar Al contacts with a width of 4 mm and a separation of 1 mm, in vacuum, at ambient temperature. 2.4. Reliability analyses
Fig. 1. Overall outline of the synthesis.

Additionally, in order to investigate the inuence of further annealing in inert atmosphere on the electrical properties of zinc oxide, some undoped lm samples were subjected to a forming gas process (FGP) performed at 500550 C for 1 h. The gas mixture employed contained 8% H2 in N2. The samples were also allowed to cool down in this atmosphere.

2.3. Characterisation A Philips (PW3719) X'pert Materials Research X-ray diffractometer, operated at 40 kV and 35 mA, with a Cu-K radiation source, was employed for XRD. All diffractograms were applied a smoothing process of 10-point Fast Fourier transform followed by an appropriate baseline correction. Scherrer's equation was used to estimate the average crystallite sizes, where the value of was taken as 0.154 nm.

In order to asses the physical reliability of the obtained lms, two separate tests were devised: Firstly, a temperature cycling test was carried out employing a Sun Electronic Systems EC11 Chamber. The temperature range and number of cycle applied were 4065 C and 100, respectively. The specimens were kept at these temperatures for 5 min, between which a heating or cooling rate of 10 C min 1 was applied. Secondly, an accelerated delamination test was performed in a laboratory set-up, a schematic representation of which is shown in Fig. 3. The size of each specimen subjected to this test was ~ 9 cm2. The applied bias and temperature were 100 VDC (10 min) and ~200 C, respectively. A Keithley 237 High Voltage Source Measure Unit was employed as the voltage source, and for reading the electric current passing through the system. Following the treatment, the specimens were immediately taken out of the set-up, and allowed to cool down in ambient atmosphere (with an approximate relative humidity of 8090%). The specimens were then observed for indications of delamination. 3. Results and discussion 3.1. Comparative characterisation 3.1.1. Sem Top-view ((a) and (b)) and cross-sectional ((c) and (d)) SEM images of the undoped ((a) and (c)) and Al-doped ((b) and (d)) lms are shown in Fig. 4. The top-view SEM images showed a welldeveloped grain structures for undoped lms. However, denser and less porous lms with smaller visual crystallite sizes were obtained upon Al doping, as anticipated [6,7]. Cross-sectional SEM images also conrmed that both types of lms could be obtained crack-free. The

Fig. 2. Outline of the lm preparation and processing.

Fig. 3. Schematic representation of the set-up used in the accelerated delamination test.

M. opurolu et al. / Thin Solid Films 517 (2009) 63236326

6325

Fig. 6. XRD patterns of the undoped and the Al-doped lms.

Fig. 4. Top-view ((a) and (b)) and cross-sectional ((c) and (d)) SEM images of the undoped ((a) and (c)) and the Al-doped ((b) and (d)) lms.

thickness values were 92 and 113 nm for the undoped and Al-doped lms, respectively. 3.1.2. UV/vis spectrophotometry The fact that both the undoped and the Al-doped lms had a similar thickness enabled the direct comparison of UV/vis spectroscopy results obtained from these lms. Fig. 5 shows the UV/vis

spectra (a) and their rst derivatives (b) of the undoped and the Aldoped lms. Both types of lms had a high transparency (N90% transmittance) in the visible region. Although there was no signicant difference between the lms in the region between ~ 550 and 800 nm, an increase in transmittance was observed in the region between 300 and ~550 nm upon Al doping, which was consistent with the relevant literature [6,7]. The SEM results discussed above also corroborated this observation since denser and less porous lms would reect the light to a lower extent, which in turn lead to a higher transparency [7]. The values of the rst derivative maximum and calculated band gap energy were 376 nm and 3.30 eV, respectively, for both lm types. This result indicated that 5% Al doping was not adequate to inuence the electronic structure of the zinc oxide material system obtained in this study. 3.1.3. Xrd XRD patterns of the undoped and the Al-doped lms are presented in Fig. 6. c-Axially preferred crystal structure was obtained in both cases. (002) was found to be 0.62 for the undoped lm. The (100) and (101) crystal orientations did not form upon Al doping; the only observable peak in the displayed region was the one corresponding to the (002) crystal orientation, whose preference thence reached its maximum ((002) = 1). The average crystallite sizes of the lms were found to be 37 and 24 nm for the undoped and the Al-doped lms, respectively. This result was consistent with the SEM observations (Fig. 4). 3.1.4. Electrical measurements Current (I) versus voltage (V) plots of the undoped lm before and after FGP, and that of the Al-doped lm are given in Fig. 7; and the minimum electrical resistivity values were found to be 4.5, 0.14 and 0.34 m, respectively. Signicant improvement in the electrical conductivity upon the application of FGP was most likely due to oxygen desorption from the grain boundaries [12] during the process. Incorporation of Al into the system also improved the electrical

Fig. 5. UV/vis spectra (a) and their rst derivatives (b) of the undoped and the Al-doped lms.

Fig. 7. IV plots of the undoped lm before and after FGP, and that of the Al-doped lm.

6326

M. opurolu et al. / Thin Solid Films 517 (2009) 63236326

then facilitate the delamination process. Despite these facts, however, none of the specimens obtained in this study exhibited delamination, which was inspected visually, upon the accelerated delamination treatment, even after days. This result, therefore, along with that of the temperature cycling test, demonstrated the high physical reliability of the lms obtained in this study. 4. Conclusions Undoped zinc oxide thin lms were obtained with desired properties. Al doping reduced the average crystallite size, and led to a denser and less porous morphology, and also caused an increase in transparency in the UV region. An improvement in electrical conductivity was achieved upon Al doping. The results of both temperature cycling and accelerated delamination tests demonstrated the high physical reliability, which was of signicant importance in TCO applications, of the thin lms of the material system obtained in this study. Acknowledgements The authors would like to acknowledge Mr. Dermot Houston for his assistance in the accelerated delamination test. References
[1] T. Minami, Semicond. Sci. Technol. 20 (2005) 35. [2] S. O'Brien, L.H.K. Koh, M. opurolu, G.M. Crean, D.P. Norton, C. Jagadish, I. Buyanova, G.-C. Yi, Zinc Oxide and Related Materials, Boston, MA, USA, 2630 November 2007, Materials Research Society Symposium Proceedings 1035E, 2008, pp. L0509. [3] S. O'Brien, L.H.K. Koh, G.M. Crean, Thin Solid Films 516 (2008) 1391. [4] L. Spanhel, M.A. Anderson, J. Am. Chem. Soc. 113 (1991) 2826. [5] L. Znaidi, G.J.A.A.S. Illia, S. Benyahia, C. Sanchez, A.V. Kanaev, Thin Solid Films 428 (2003) 257. [6] H.-M. Zhou, D.-Q. Yi, Z.-M. Yu, L.-R. Xiao, J. Li, Thin Solid Films 515 (2007) 6909. [7] A.Y. Oral, Z.B. Bahi, M.H. Aslan, Appl. Surf. Sci. 253 (2007) 4593. [8] T. Schuler, M.A. Aegerter, Thin Solid Films 351 (1999) 125. [9] R.B.H. Tahar, J. Eur. Ceram. Soc. 25 (2005) 3301. [10] Bellcore, Generic Reliability Assurance Requirements for Optoelectronic Devices Used in Telecommunications Equipment (GR-468-CORE), Issue 1, Bellcore, 1998. [11] K.W. Jansen, A.E. Delahoy, Thin Solid Films 423 (2003) 153. [12] V. Musat, B. Teixeira, E. Fortunato, R.C.C. Monteiro, Thin Solid Films 502 (2006) 219.

Fig. 8. Cross-sectional SEM images of the undoped (a) and the Al-doped (b) lms after the temperature cycling test.

conductivity. This was expected as Al atoms would act as donors and therefore contribute to the electrical conductivity. Nonetheless, it was obvious that further investigation was necessary in order to further improve the electrical conductivity as resistivities lower than those obtained in this study could be found in the literature [6,8,9]. 3.2. Reliability analyses 3.2.1. Temperature cycling test Temperature cycling test was performed for both the undoped and the Al-doped lms. As seen from the cross-sectional SEM images (Fig. 8(a) and (b), respectively), no crack formation was observed in either of the specimens after the temperature cycling treatment. 3.2.2. Accelerated delamination test Accelerated delamination test was performed for both the undoped lm (before and after FGP), and the Al-doped lm. During the application of the voltage, an electric current of a magnitude of 23 A was observed passing through the set-up, which then decreased gradually down to several hundred nA, in all cases. Therefore, cations (mainly Na+) must have been forced to migrate through the glass substrate [11]. This intentional driving of cations was anticipated to indirectly cause the TCO to delaminate from the glass substrate through some chemical reactions proposed previously [11]. Also, the physical presence of high concentrations of sodium could lead to increased mechanical stress at the TCOglass substrate interface, which might